XCENA has introduced its MX1 production memory lineup, designed to address the growing memory constraints in AI infrastructure. The MX1 lineup aims to overcome limitations in capacity, utilization, and data movement, which are currently hindering AI inference at scale. As AI models become larger and context windows expand, memory emerges as a critical bottleneck. XCENA's CXL-based MX1 solutions allow memory to scale efficiently alongside compute, offering an alternative to costly High-Bandwidth Memory and the inefficiencies of scaling out with additional servers.

"AI performance is no longer limited by compute, it's limited by memory, and MX1 is our answer," stated Jin Kim, CEO of XCENA. The MX1 lineup includes MX1 Compute, which pairs CXL-based memory expansion with near-data processing using 2,048 RISC-V cores, and MX1 Expand, offering eight DRAM slots to support DRAM reuse and server scale-up. These solutions are being showcased at FMS 2026.

XCENA is collaborating with Intel Xeon 6 platforms to demonstrate how CXL-based memory can effectively manage memory-intensive AI workloads. The company is advancing proof-of-concept engagements toward production-level evaluation and commercialization with hyperscalers, cloud service providers, and enterprise AI infrastructure customers. XCENA previously received recognition at FMS for its innovative memory technologies.